Participants provided 637 measurement days before and 557 days after retirement. work-related physical activity was defined as physical activity accumulated at workplace. commuting activity was dichotomised based on the speed of trips between home and workplace to active (<20 km/h) and passive ( 20 km/h) commute. participants were divided into four groups: non-active workers and commuters, non-active workers but active commuters, active workers but non-active commuters, and active workers and commuters. linear regression analysis with generalized estimating equations were used for statistical analysis. results: the change in physical activity during retirement transition markedly varied by the activity group. lower work-related activity was associated with an increase in light physical activity and a decrease in sedentary time. conversely, higher work-related activity was associated with a decrease in light physical activity and an increase in sedentary time, except among those active workers who were active commuters. particularly the active workers but non-active commuters increased their sedentary time (48 min, 95% cl 20 to 76) and decreased their light physical activity (54 min, 95% cl -80 to - 29). no statistically significant changes in moderate-to- vigorous physical activity were observed. conclusions: our results suggest that work-related physical activity is associated with changes in physical activity behavior when retiring. special attention should be targeted to active workers who are non-active commuters to maintain physical activity and decrease sedentary time after retirement. key messages: lower work-related physical activity before retirement was associated with an increase in light physical activity and a decrease in sedentary time after retirement. active workers but non-active commuters showed an increase in sedentary time and a decline in light physical activity after retirement, and might be a suitable group for physical activity promotion